1 files changed, 117 insertions, 0 deletions
diff --git a/app/scrapers/marz-shinjuku.ts b/app/scrapers/marz-shinjuku.ts
new file mode 100644
index 0000000..06c0bcc
--- /dev/null
+++ b/app/scrapers/marz-shinjuku.ts
@@ -0,0 +1,117 @@
+/**
+ * 新宿MARZ — http://www.marz.jp/schedule/YYYY/MM/
+ *
+ * DOM構造:
+ *   <article id="eXXXX">
+ *     <a href="detail-url"><div class="img"><img src="..."><time datetime="YYYY-MM-DD"></div></a>
+ *     <h1><a href="...">タイトル</a></h1>
+ *     <div class="desc">
+ *       <div class="entrybody"><p>【LIVE】<br/>アーティスト名</p> ...</div>
+ *       <div class="entryex"><p>open HH:MM / start HH:MM<br/>adv¥XXX / door¥XXX<br/>...</p></div>
+ *     </div>
+ *   </article>
+ */
+import * as cheerio from "cheerio";
+import type { Scraper, VenueMeta } from "./base";
+import type { EventInput } from "~/lib/db.server";
+
+export const venue: VenueMeta = {
+  id: "marz-shinjuku",
+  name: "新宿MARZ",
+  url: "http://www.marz.jp",
+  area: "新宿",
+  capacity: 200,
+};
+
+async function scrapeMonth(year: number, month: number): Promise<EventInput[]> {
+  const url = `http://www.marz.jp/schedule/${year}/${String(month).padStart(2, "0")}/`;
+  const res = await fetch(url);
+  if (!res.ok) throw new Error(`HTTP ${res.status} for ${url}`);
+  const $ = cheerio.load(await res.text());
+  const events: EventInput[] = [];
+
+  $("article").each((_, el) => {
+    const $el = $(el);
+
+    const date = $el.find("time[datetime]").attr("datetime");
+    if (!date) return;
+
+    const title = $el.find("h1 a").first().text().trim();
+    if (!title) return;
+
+    const source_url = $el.find("a[href]").first().attr("href") ?? null;
+    const image_url = $el.find(".img img").attr("src") ?? null;
+    const ticket_url =
+      $el
+        .find(".entrybody a[href]")
+        .filter((_, a) => {
+          const h = $(a).attr("href") ?? "";
+          return h.startsWith("http") && !h.includes("marz.jp");
+        })
+        .first()
+        .attr("href") ?? null;
+
+    // Artist lines: clone entrybody, replace <br> with \n, filter out labels/ticket lines
+    const $body = $el.find(".entrybody").clone();
+    $body.find("br").replaceWith("\n");
+    const artistLines = $body
+      .text()
+      .split("\n")
+      .map((l) => l.trim())
+      .filter(
+        (l) => l && !l.includes("【") && !l.startsWith("■") && !l.startsWith("※")
+      );
+    const artist = artistLines.length > 0 ? artistLines.join(" / ") : null;
+
+    // Time and price from entryex
+    const $ex = $el.find(".entryex").clone();
+    $ex.find("br").replaceWith("\n");
+    const exLines = $ex
+      .text()
+      .split("\n")
+      .map((l) => l.trim())
+      .filter(Boolean);
+
+    const timeLine = exLines[0] ?? "";
+    const openMatch = timeLine.match(/open\s+(\d{1,2}:\d{2})/i);
+    const startMatch = timeLine.match(/start\s+(\d{1,2}:\d{2})/i);
+    const priceLine =
+      exLines.find((l, i) => i > 0 && /[¥￥]|adv|door/i.test(l)) ?? null;
+
+    events.push({
+      venue_id: venue.id,
+      title,
+      artist,
+      date,
+      open_time: openMatch?.[1] ?? null,
+      start_time: startMatch?.[1] ?? null,
+      price: priceLine,
+      ticket_url,
+      image_url,
+      source_url,
+    });
+  });
+
+  return events;
+}
+
+export const scraper: Scraper = {
+  venue,
+  async scrape(): Promise<EventInput[]> {
+    const now = new Date();
+    const next = new Date(now.getFullYear(), now.getMonth() + 1, 1);
+
+    const [thisMonth, nextMonth] = await Promise.all([
+      scrapeMonth(now.getFullYear(), now.getMonth() + 1),
+      scrapeMonth(next.getFullYear(), next.getMonth() + 1),
+    ]);
+
+    const seen = new Set<string>();
+    return [...thisMonth, ...nextMonth].filter((e) => {
+      const key = `${e.date}|${e.title}`;
+      if (seen.has(key)) return false;
+      seen.add(key);
+      return true;
+    });
+  },
+};